#!/usr/bin/env bash

lines=$(cat dril.html |\
    # war crimes
    grep "<div class='text'>" | sed -e "s%<div class='text'>\(.*\)</div>%\1%" |\
    # remove links
    grep -vE 'https?://' |\
    # remove RTs and replies (but not inline @'s)
    grep -v -e '^RT @' -e '^@' |\
    # unescape html entities
    sed -e 's/\&amp;/\&/g' -e 's/&lt;/</g' -e 's/&gt;/>/g'
)
total=$(echo -n "$lines" | wc -l)
at=1
while read tweet; do
    echo "$tweet" | sed 's%<br />%\n%g'
    echo "                -- dril"
    echo %

    (( at++ ))
    >&2 echo -ne "\rProcessed: $at / $total"
done <<< "$lines"